NVIDIA H200 SXM Specs

NVIDIA H200 SXM
16896 Shaders
141GB HBM3e
1980MHz
Peak Performance
3957.72 TOPS
INT8 Tensor Sparse
FP32
66.91 TFLOPS
FP16
133.82 TFLOPS
Form Factor
SXM5
-
TDP
700W
-
-
-
-
-
GB6 OpenCL N/A
0%
GB6 Metal N/A
0%
GB6 Vulkan N/A
0%
GB5 OpenCL N/A
0%
GB5 CUDA N/A
0%
GB5 Metal N/A
0%
GB5 Vulkan N/A
0%
OCT 2020.1 N/A
0%
OCT Metal N/A
0%
Manufacturer
NVIDIA
Chip Designer
NVIDIA
Architecture
Hopper
Family
Server
Codename
NV180
GH100
-
-
Market Segment
Server
Release Date
11/13/2023
Foundry
TSMC
-
Fabrication Node
4N
-
Die Size
814 mm²
-
Transistor Count
80 Billion
-
Transistor Density
98.28M/mm²
-
Form
SXM5
Shading Units
16896 Shaders
-
Texture Mapping Units
528 TMUs
Render Output Units
24 ROPs
Tensor Cores
528 T-Cores
Ray-Tracing Cores
704 RT-Cores
Streaming Multiprocessors
132 SMs
-
-
-
-
-
-
1830MHz Tensor
-
1065MHz Base
1980MHz
Peak Performance
3957.72 TOPS
INT8 Tensor Sparse
-
-
-
FP8
-
1978.86 TFLOPS Tensor (FP16 Accumulate)
3957.72 TFLOPS Tensor (FP16 Accumulate) Sparse
1978.86 TFLOPS Tensor (FP32 Accumulate)
3957.72 TFLOPS Tensor (FP32 Accumulate) Sparse
FP16
133.82 TFLOPS
989.43 TFLOPS Tensor (FP16 Accumulate)
1978.86 TFLOPS Tensor (FP16 Accumulate) Sparse
989.43 TFLOPS Tensor (FP32 Accumulate)
1978.86 TFLOPS Tensor (FP32 Accumulate) Sparse
FP32
66.91 TFLOPS
-
-
FP64
33.45 TFLOPS
61.84 TFLOPS Tensor
BF16
133.82 TFLOPS
989.43 TFLOPS Tensor
1978.86 TFLOPS Tensor Sparse
TF32
494.72 TFLOPS Tensor
989.43 TFLOPS Tensor Sparse
-
-
-
INT8
-
1978.86 TOPS Tensor
3957.72 TOPS Tensor Sparse
INT32
33.45 TOPS
-
-
Pixel Fillrate
47.52 GPixel/s
-
-
Texture Fillrate
1045.44 GTexel/s
-
-
L1
64KB/SM Tex
256KB/SM
-
-
L2
50MB Shared
-
-
-
141GB
HBM3e
-
Bus Width
6144Bit
Clock
1566MHz
Transfer Rate
6.3GT/s
Bandwidth
4810.8GB/s
-
-
-
-
-
-
-
-
-
TDP
700W
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
No Ports
Max Resolution
Unknown
Max Resolution Refresh Rate
-
Variable Refresh Rate
G-Sync
FreeSync
-
Display Stream Compression (DSC)
Not Supported
Multi Monitor Support
Unknown
-
-
No Encoders
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
No Decoders
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
OpenCL
3.0
-
-
-
-
CUDA
9.0
-
-
PureVideo HD
VP11
VDPAU
Feature Set K
Not a Card
-
-
-
-
-
-
-
-
-
-
-
-
-
PCIe Version
5.0
PCIe Lanes
16
Multi GPU Support
Supported
Type
NVLink
-
-
-
-
-
-
Change Comparison